NAIC Life RBC C-3 Field Test Targets New GOES Generator: What Changes for VA and FIA Capital

The NAIC Life Risk-Based Capital (E) Working Group confirmed at the March 2026 Spring National Meeting that a Summer 2026 C-3 field test will run on the new Generator of Economic Scenarios, in parallel with the legacy Academy Interest Rate Generator, with year-end 2027 targeted for VM-20 and VM-21 adoption and C-3 RBC following on the same calendar or a one-year lag.

Comment deadlines on the separate CLO RBC factor proposal closed April 16 and 17. Two components of life RBC are therefore moving at once, and the C-3 change is decided by tail shape rather than by any formula amendment.

Key Takeaways

The field test is a parallel run: participants compute C-3 on a representative block under both AIRG and GOES and submit the difference with attribution and commentary.
AIRG was calibrated to Treasury data through 2004 and S&P 500 returns over 1955 to 2003, so its paths above the 95th percentile cluster on persistent low-rate environments.
Because C-3 Phase II uses CTE95, the shape of the distribution above the 95th percentile is the capital number, and GOES puts significant mass on sustained high-inflation, high-rate paths.
Running both generators over a representative GMWB block, programs hedging mainly with equity index options and limited rate hedging see C-3 increases of 5% to 12%, while programs with robust rate hedges show smaller or slightly negative impacts.
Comment deadlines of April 16 and 17 on CLO factors move the C-1 component for the same carriers, with both work streams likely to produce final guidance inside the same 12-month window.

What Is Being Tested

GOES is the Academy and Society of Actuaries project replacing AIRG and its companion equity scenario set with a single integrated multi-asset stochastic model. AIRG was calibrated to US Treasury yield data through 2004 and reflects the regime that produced it: stable inflation, declining long rates, modest equity volatility. The post-2008 zero rate period, the 2022 inflation shock and the back-up in long-end yields between 2022 and 2024 all sit outside the conditions it was built to represent.

The field test is structured as a parallel run. Participating companies calculate C-3 capital on a representative block under both generators, submit the results, and comment on operational and capital impact. The Working Group session on March 24 confirmed the timing, accepted the most recent GOES calibration update and aligned on the submission window, which is expected to close in late August or early September with review through the Fall National Meeting.

Its output decides three open questions: whether the GOES calibration should be modified before adoption, whether the C-3 formula factors need re-anchoring to the new generator, and whether transition relief is needed where capital shifts materially.

Five design differences drive that:

Cascade rather than sequential generation. AIRG builds rate paths first and conditions equity returns on them. GOES drives both from common macroeconomic factors, producing a coherent joint distribution and a richer set of inflation-driven scenarios.
Real and nominal decomposition. GOES separates real rates from breakeven inflation and recombines them, so high realized inflation can produce high nominal rates without the sustained low-rate paths AIRG's mean reversion requires.
Equity calibration window. AIRG's regime-switching equity model was fitted to 1955 through 2003. GOES extends through 2024 and shows fatter left tails at long horizons, partly from including the 2008 to 2009 and 2020 drawdowns.
Softer mean reversion. Extreme rate environments persist longer before reverting, which matters over the 30 plus year liability horizons typical of annuity blocks.
Tail composition. AIRG paths above the 95th percentile cluster on persistent low rates. GOES carries mass in both directions.

The Metric Is CTE95, So the Tail Is the Capital

C-3 Phase II computes stochastic capital as a conditional tail expectation at the 95th percentile of the present value of greatest accumulated deficit. Nothing below that point enters the number. A generator change that leaves the body of the distribution alone and rewrites its top five percent therefore changes capital directly, with no amendment to the formula.

That is what the AIRG calibration window produced. Fitted to data through 2004, its extreme paths are extreme in one direction, and sustained high-rate environments are underrepresented at exactly the percentile the metric reads. GOES puts prolonged high-inflation, high-rate paths into that region alongside the low-rate ones, which changes the mix of in-the-money and out-of-the-money outcomes for a GMWB rider liability.

The direction depends on the hedge program, not the block. High-rate tail paths compress rider value, because guaranteed cash flows are discounted more heavily, while the same paths lengthen high-volatility periods and increase hedge slippage. Running both generators over a representative GMWB block, programs hedging primarily through equity index options with limited rate hedging see C-3 increases of 5% to 12% depending on block characteristics; programs maintaining robust rate hedges with active rebalancing show smaller or slightly negative impacts.

Correlation is the second lever. Hedging built on the inverse rate-equity relationship that held from 1990 to 2020 carries higher residual risk in GOES scenarios where rates and equities move together, which raises gross hedge cost over the projection horizon. Behavior assumptions compound it: lapse and partial-withdrawal assumptions calibrated against AIRG scenarios trigger in a different proportion of GOES paths, so the recalibrated assumption set moves the liability profile before any capital calculation runs.

The Evidence Base Is Assembled From Self-Selected Blocks

The field test does not require the full in-force. Companies submit a representative block capturing material C-3 exposure, and those that cannot complete the package may participate at reduced scope with explicit identification of what was not modeled and why.

Both concessions are operationally necessary, given the lift of running two generators through production VM-20 and VM-21 pipelines. They also mean the aggregate the Working Group calibrates against is built from blocks each participant chose, at a scope each participant set, by companies that already know roughly which direction the change moves their capital. Confidentiality protections cover individual submissions and only aggregates may be published, which supports participation, and it also means the selection cannot be inspected from outside.

Operational readiness constrains the same population. Many carriers run VM-20 and VM-21 production on legacy actuarial software that needs vendor releases supporting GOES before AIRG can be retired, and production-quality releases covering the full calculation set are not yet broadly available. If those slip, year-end 2027 becomes a second year of dual running, with scenario library generation, stochastic-on-stochastic projection for VM-21 and attribution reconciliation all doubled.

Meanwhile the other half of the ratio is moving. The CLO factor recalibration would shift the C-1 asset risk component toward look-through factors based on underlying loan exposure rather than NAIC designation, with higher charges likely on lower-rated and structurally subordinated tranches. CLO holdings have grown materially in life general accounts over the past decade, concentrated at private equity affiliated carriers and reinsurers, which is much the same population running large annuity blocks through the C-3 field test. A company modeling the GOES impact on its capital ratio while holding C-1 fixed is measuring one leg of a change that arrives on both.

NAIC Life RBC C-3 Field Test Targets New GOES Generator: What Changes for VA and FIA Capital

Key Takeaways

What Is Being Tested

The Metric Is CTE95, So the Tail Is the Capital

The Evidence Base Is Assembled From Self-Selected Blocks

Further Reading

Sources